Вы находитесь на странице: 1из 25

J. Child Lang. 31 (2004), 287310. f 2004 Cambridge University Press DOI: 10.

1017/S0305000904006002 Printed in the United Kingdom

A longitudinal study of language development in two children with Williams syndrome*


YONATA LEVY The Hebrew University
(Received 7 May 2002. Revised 23 May 2003)

ABSTRACT

Williams syndrome (WS) is often cited as the prime example within developmental disorders of the dissociation of language from other cognitive skills, particularly from visuo-motor skills. This claim has been responsible for the challenges posed by this population to cognitive theories and to models of language acquisition. Two Hebrew-speaking children with WS were followed longitudinally for 18 months, from the rst occurrences of two word combinations. Y was 3 ; 95 ; 3 and BT was 4; 25 ; 8 during data collection. Both children had an MLU of 1.82.8. The childrens linguistic proles were compared to the proles of typically-developing (TD) children of similar MLU as well as to the proles of 11 children with a variety of neurodevelopmental disorders (ND), matched on MLU. The proles exhibited by the children with WS throughout the period of the study diered from the proles seen in both control groups.

INTRODUCTION

Williams syndrome (WS; Williams, Barratt-Boyes & Lowe, 1961) is a rare autosomal genetic disorder (1 : 25 000 live births) characterized by typical facial dysmorphology, renal and cardiovascular anomalies, statural deciencies, characteristic dental malformation and hypercalcemia (McKusick, 1988). Most cases of WS are non-familial (but see recent ndings, Osborne, Pober, Chitayat, Bodurtha, Mandel, Costa, Grebe, Cox, Tsui & Scherer, 2001). A microdeletion on chromosome 7q11.23 has been identied in 98% of the individuals with WS. The missing region typically includes the ELN gene, which is hypothesized to account for the vascular and connective tissue abnormalities. The other phenotypic characteristics are presumably
[*] This work was supported by the Israeli Science Foundation, Grant 795/97. I am grateful to the Israeli Williams Syndrome Association for their help and enthusiasm. Address for correspondence : Yonata Levy, Psychology Department, The Hebrew University, Jerusalem, Israel 91905. fax : 972-2-5881159. e-mail : msyonata@mscc.huji.ac.il

287

LEVY

linked to the adjacent 16 or more genes that are part of the standard deletion in WS. IQ in individuals with WS is typically within the 5070 range. Reliable linkages between specic gene deletions and details of the cognitive prole of WS have not yet been found (Mervis, Morris, Bertrand & Robinson, 1999 ; Tassabehji, Metcalfe, Karmilo-Smith, Carette, Grant, Dennis, Reardon, Splitt, Read & Donnai, 1999). The cognitive prole of individuals with WS has been characteristically described as having spared auditory short term memory and spared linguistic and face recognition abilities, in the face of serious decits in number, visuo-spatial cognition, motor behaviour, planning and problem solving (Bellugi, Marks, Bihrle & Sabo, 1993; Gosch, Stading & Pankau, 1994 ; Udwin & Dennis, 1995; Mervis et al., 1999). A distinct behavioural phenotype and a relatively well-understood genotype have made WS a particularly good candidate for brain-behaviour studies and as such it has attracted the interest of researchers in a number of domains. WS is often cited as the prime example within developmental disorders, of the dissociation of language from other cognitive skills, particularly from visuo-motor skills. In recent years research has focused on the exact nature of the linguistic competence that individuals with WS exhibit. This work resulted in controversial positions, with an increasing number of studies casting doubt on the intactness of linguistic competence in individuals with WS. At the focus of the current debate are studies of morpho-syntax and lexical semantics. In the review of the literature below we focus on the former since this is the area of linguistic knowledge most relevant to the current study. In her seminal work on the cognitive prole of adolescents and adults with WS, Bellugi and colleagues (e.g. Bellugi, Bihrle, Neville, Doherty & Jernigan, 1992 ; Bellugi et al., 1993) argued that the linguistic prole of individuals with WS shows a sparing of syntax both in comprehension and in production, which extends to tests of metalinguistic abilities as well. Morphological markers are generally used correctly, including markers for tense and aspect, and so are auxiliaries and articles (Bellugi et al., 1993). In most of these studies participants with WS were school age children or adolescents and their achievements were compared to those of individuals with Down syndrome. Udwin & Dennis (1995) are in agreement with the view expressed by Bellugi and colleagues. They too conclude that mature individuals with WS have an unusual command of language. Their comprehension is usually far more limited than their expressive language, which tends to be grammatically correct, complex and uent at a supercial level but verbose and pseudo-mature in its content. Mervis et al. (1999) argued that a unique cognitive prole characterized individuals with WS. Contrary to previous work (Bellugi et al., 1992, 1993) 288

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT

they concluded that the language abilities of individuals with WS as measured on standardized tests were signicantly delayed relative to CA controls. Delay was evident on receptive vocabulary, receptive grammar, semantic uency and syntactic measures (see also Volterra, Capirci, Pezzini, Sabbadini & Vicari, 1996; Karmilo-Smith, Grant, Berthoud, Davies, Howlin & Udwin, 1997). Syntactic abilities as well as MLU were at the expected level for mental age (MA). Furthermore, while achievements on visuo-spatial tasks were signicantly lower than scores on verbal tasks, a correlation was found between achievements on these two types of tasks, suggesting that language in individuals with WS was not independent of other cognitive skills. Studies of Italian-speaking individuals with WS are of particular relevance to the current work since Italian, like Hebrew, has complex morphology. Pezzini, Vicari, Volterra, Milani & Ossella (1999) studied the cognitive prole of Italian children with WS. Children with WS ages 4; 1015 ; 3, with a mean MA of 5; 4 were compared to TD children ages 3; 87 ; 6 (mean age 5 ; 6). Within the linguistic domain, children with WS performed better than their MA matched controls on phonological uency. They were poorer, however, on the Boston Naming Test. There were no group dierences on any of the other language tasks. As for the perceptualmotor tasks, children with WS performed better than their MA controls on face recognition and signicantly poorer on the WISC-R block design. There were no signicant dierences on any of the other perceptual-motor tasks (see also Volterra et al., 1996). Interestingly, an examination of individual proles in the Pezzini et al. (1999) study revealed a large variability in performance among the participants with WS including on variables that distinguished the group as a whole from the typically-developing children. The Italian study argues against Mervis et al. (1999) who report a characteristic cognitive prole of individuals with WS. However, in a recent study of the cognitive prole of Hebrew-speaking adolescents with WS we replicated Mervis et al. (1999), both with respect to the group as a whole and with respect to individual proles (Levy & Bechar, 2003). There was no evidence in the Hebrew data of the individual variability seen in the Italian group. Note that Pezzini et al. (1999) and Volterra et al. (1996) are open to the criticism raised by Mervis & Robinson (1999) concerning the need to control for chronological age as well as for mental age when comparing across subjects and across tasks. Summarizing, performance of individuals with WS on standardized tests yields an advantage of performance on verbal tasks over other tasks, specically over visuo-motor tasks. Yet, achievement on verbal tasks is not superior in any sense. It varies considerably among individuals with WS and typically is at MA level. Auditory memory for words seems preserved and it may even reach the expected CA level. 289

LEVY

Along with assessments on standardized tests, individuals with WS have been tested on experimental, non-standardized tasks. Such tasks are particularly important when claims concerning preserved linguistic abilities are at stake. One of the central current debates concerns morpho-syntax in English speaking individuals with WS, specically, plurals and past tense formation. Clahsen & Almazan (1998, 2001) and Clahsen & Temple (2003) focused on morphology and morpho-syntax in children with WS. Past tense formation, noun plurals and compounding were investigated in four English-speaking children with WS. Results showed that children with WS over-regularize the past tense /-ed/, and the plural /-s/ signicantly more often than MA controls, whereas knowledge of irregular past and irregular plural was relatively poor. The children over applied the /-s/ plural to the internal nominal elements within lexical compounds (yielding, for example, the ungrammatical compound *rats-eater instead of rat-eaters ). Similarly, the children over applied the regular comparative ax /-er/ where a comparative more was required. Clahsen and colleagues interpret these ndings as evidence of a preserved grammatical rule system, along with considerable decits in lexical knowledge such as is implicated in knowledge of irregular forms. Zukowsky (2001) too found a regular/irregular asymmetry in plural production in children with WS. The participants resisted over-regularizations of plural /-s/ to irregulars and were eventually able to produce the irregular forms. Consequently, their decits were interpreted as evidence for problems in retrieval. Thomas, Grant, Gsodl, Laing, Barham, Lakusta, Tyler, Grice, Patterson & Karmilo-Smith (2001) examined past tense formation in a group of 18 English-speaking individuals with WS. Contrary to Clahsen and colleagues, they found no selective irregular verb decit once dierences in mental age were controlled for. The study revealed a decit in generalization of past tense regularities to novel forms in the WS group. In a study of Hebrew morpho-syntax, 10 participants with WS were compared to TD children (Levy & Hermon, 2003). Given the variability in IQ scores that characterized the WS group, averaging over participants IQ was not attempted. Instead, participants were compared to two control groups : a younger group age 5; 7 (5 ; 35 ; 11), (which spanned the lower end of the WS IQ scores), and an older group age 11;7 (10 ;312;6), (which was at the upper end of the WS IQ level). The participants with WS were signicantly worse than the older control group on most sub-tests, that is, they did not score higher than their average-to-low MA. Several studies considered the early phases of language development in toddlers with WS. As far as precursors to language were concerned, a reduction in pointing and impairments in triadic interactions were found (Laing, Butterworth, Ansari, Gsodl, Longhi, Panagiotaki, Paterson 290

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT

& Karmilo-Smith, 2002). Similar to all other populations with neurodevelopmental disorders, the onset of language is usually delayed in WS. Singer-Harris, Bellugi, Bates, Jones & Rossen (1997) suggested that the mean delay was of two years. Vocabulary growth in WS was not correlated with joint attention (Mervis & Bertrand, 1997). In a preferential looking task, toddlers with WS did not outperform toddlers with Down syndrome in matching words to pictures. Both groups performed at MA level and were signicantly worse than CA normal controls (Paterson, Brown, Gsodl, Johnson & Karmilo-Smith, 1999). As for expressive vocabulary, Mervis & Robinson (2000) reported of an advantage of two-year-old toddlers with WS over age-matched controls with Down syndrome. The expressive vocabulary advantage was present even before any of the children began to combine words. Relatively little is known about the acquisitional course of language in young children with WS. To the best of my knowledge, Capirci, Sabbadini & Volterra (1996) is the only published longitudinal follow-up of language development in a child with WS. Observations of this Italian-speaking girl began when she had about 20 words and no syntax. The study reports similarities between the acquisitional course followed by this child and the developmental course observed in normal children. However, some dierences were noted as well. In particular, while the child had good vocabulary and procient syntax she made agreement errors and errors of pronominalization of the kind not documented in normally developing Italian children. By the age of 4; 10, the childs cognitive prole as measured on normative tests, was typical of WS, with linguistic abilities better than visuo-spatial ones, although the latter were less impaired than one often sees in children with WS. What might be the contribution of a documentation of the early stages of language acquisition in WS to the theoretical debate surrounding the linguistic abilities of individuals with this syndrome ? Most of the work that concerned mature linguistic competence in disordered populations has not considered it theoretically necessary to relate the observations to the developmental trajectories en route to mature competence. It may be argued, however, that if there were normal onset and normal rate and course of acquisition the claim for the relative sparing of language in individuals with WS in the face of general cognitive decits would be greatly enhanced. Karmilo-Smith and colleagues (Karmilo-Smith, 1998 ; KarmiloSmith, Scerif & Thomas, 2002), however, argue that an investigation of the course of acquisition is bound to reveal atypical pathways. Thus, even if there are behaviourally intact cognitive systems in the mature individual, normalcy is only apparent and a ne-grained analysis is likely to discover that the underlying knowledge base is in fact atypical. This is so since the eects of anomalous genotypes are indirect and non-specic and likely to 291

LEVY

aect internal brain environment and the ways in which that environment interacts with the input to achieve learning (Karmilo-Smith, 1998). Whether atypical brains in fact follow atypical developmental trajectories is an empirical question. It is still possible that despite the disruption of normal brain structures, compensatory mechanisms may provide for normal learning.1 Note that some of the studies summarized above bear upon these issues. Data is still scant, however, with respect to early grammatical development in WS. The current work reports on a longitudinal follow up of two Hebrewspeaking children with WS. The linguistic proles of the children with WS were compared to proles of typically-developing (TD) children as well as to children with other neurodevelopmental disorders of similar MLU (Levy, Tennebaum & Ornoy, 2000, summarized in Table 3). Although we used our own control group, the linguistic variables that were studied were mostly those that have been investigated in TD Hebrew-speaking children. Below I give a short summary of the morphological and syntactic phenomena in Hebrew that are relevant to the current study. Related acquisitional facts, describing typical development are summarized as well. A brief description of relevant aspects of Hebrew grammar and related acquisitional facts Hebrew has the characteristics of Semitic languages, i.e. words are composed of consonantal roots cast in vocalic word patterns. The roots are usually tri-consonantal while the patterns are in the form of vocalic inxes, prexes and suxes. There are seven verb patterns binyanim and about three-dozen noun patterns mishkalim. All verbs are analysable into root+pattern. With respect to nouns, however, this generalization is only partial since some nouns do not have a recognizable root. It is generally the case in Hebrew that the roots convey core meanings, while the derivational paradigms may partially introduce meaning modulations. While the formal paradigms are highly systematic and well-formedness is clearly dened, the semantics that the derivational patterns convey is only partially predictable from their forms. Hebrew has a rich inectional morphology. Generally, verbs are inected for tense, number, person and gender while nouns have gender and are marked for plurality. Agreement with respect to gender, number and person is obligatory between subjects and predicates as well as within noun phrases. Nouncomplement constructions that are without verbs, traditionally called nominal sentences , are well formed in Hebrew. An accusative
[1] For a discussion of the notion of compensatory mechanisms and brain plasticity and its application to cases of congenital disorders, see Levy, 2003.

292

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT

marker /et/ is obligatory. Footnote 2 gives examples of agreement, verbless clauses and the use of accusative /et/.2 The acquisition of Hebrew morphology has been studied extensively in recent years. Berman (1985, 1994) investigated the development of verb morphology. She argues that early verb use is rote learned and thus, item based. Once the child begins to vary verb forms, each root will appear in many patterns and with a variety of inectional endings.3 Hebrew-speaking children start out with SEMANTICALLY unanalysed forms of verbs yet at the same period they can eectively control the necessary formal manipulations of the various root+pattern combinations. It is only around age four, a long time after they have been using most verb patterns productively, that childrens errors indicate that they begin to appreciate the semantics of the system (Berman, 1985, 1994). Previous studies of the acquisition of gender in a variety of languages, including Hebrew, show that children master the formal-morphological parts of this system relatively early (Mulford & Morgan, 1983 ; Smoczynska, 1985 ; Levy, 1988). Thus, even under age 3; 0, errors of linguistic gender on inanimate nouns, which mark gender morphologically, are rare. In cases of animate nouns in which linguistic gender is determined by the semantic notion of gender, errors are common and learning is a more protracted

[2] Examples 14 illustrate agreement patterns in Hebrew. Examples 57 demonstrate noun+complement constructions and the use of accusative et. ha-xadasha 1. ha-yeled nixnas lakita the-boy entered-sg/m to(def)-class/f the-new/f The boy entered the new class lakit-ot ha- xadash-ot 2. ha-yalda nixnesa the-girl entered-sg/f to(def)-classes/f/pl the-new/f/pl The girl entered the new classes laxadarim ha-xadashim 3. ha-yeladim nixnesu the-boys entered-pl to(def)-room/m/pl the-new/m/pl The boys entered the new rooms laxadarim ha-xadashim 4. ha-yeladot nixnesu the-girls entered-pl to(def)-rooms/m the-new/pl/m The girls entered the new rooms 5. ha-yeled xaxam the-child clever The child (is) clever 6. ima babayt mother in(def)-home Mother (is) at home 7. ima raata et ha-yeled mother saw-past/f/sg (acc.marker) the(def)-boy Mother saw the boy [3] For example, the root G-D-L may appear in the childs speech in the following forms : GaDaL he grew up ; hiGDiL he increased ; GuDaL he was grown-passive ; GiDeL he grew-causative.

293

LEVY

process. These ndings hold across all languages studied so far, among them Hebrew (Levy, 1988). In a cross-sectional survey of productive syntax in Israeli children aged 1 ; 05 ; 6, Dromi & Berman (1986) found an increase in the use of polyclause utterances and a decrease in the use of one-clause utterances, documented up to age 4; 0. A decrease was observed in the production of nominal and copular clauses as well, along with an increase in the use of clauses with nite verbs. This is in line with the view that an increase in the use of sentences with nite verbs characterizes more advanced grammatical stages (Berman, 1994 ; Rice & Wexler, 1996). Previous studies have shown that at or around MLU 3, children control agreement and errors become minimal. Crucial parts of the verb morphology are likewise acquired around this stage. A reduction in errors of morphology along with a wealth of derivational and inectional forms suggests that the core structures of Hebrew morphology have been acquired at this developmental stage.4 At about the same time, use of accusative marker -/et/- is almost error free and the proportion of complex clauses increases (Berman, 1985 ; Dromi & Berman, 1986 ; Levy, 1988). However, unlike errors of morphology, the number of errors of syntax and meaning at this stage does not decrease in a similar fashion. In other words, while morphology has been largely mastered, the childs syntactic and lexical repertoires grow richer and more complex but her performance with respect to those aspects is not as yet error free (Levy et al., 2000). In sum, research in the acquisition of Hebrew as well as of other languages has uncovered developmental trajectories and in some cases suggested theoretical motivation for the simultaneous emergence of specic linguistic features in the childs repertoire. Thus, it was suggested that agreement is conditioned upon the emergence of functional categories. Most clearly, agreement is part and parcel of verb niteness (Hyams & Wexler, 1993 ; Rice & Wexler, 1996 among many). The present study does not discuss these theoretical perspectives but focuses instead on the developmental trajectories observed in toddlers with WS and the way they compare with normal development. Consequently, the variables that entered the linguistic proles presented in the current study were chosen according to the following : similar to other languages, in Hebrew, increased use of tensed verbs indicates that functional categories have developed. Less use of verbless clauses that are a common grammatical construction in Hebrew, along with increase in the use of complex clauses signal improved syntax. Gender marking on inanimate nouns is a formal morphological marker that
[4] The full derivational system of Hebrew is not mastered by children until they are well into the school years.

294

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT

is acquired prior to or around MLU 3. Agreement is a syntactic operation that children master when MLU is about 3 and the Hebrew accusative marker /et/ is likewise acquired at that time.

THE STUDY

Participants and method of data collection One girl, BT, and one boy, Y, were studied in naturalistic play situations in their respective homes. The children had a conrmed genetic diagnosis of WS (FISH ; Mervis et al., 1999). In both cases pregnancy and birth were uneventful. WS was diagnosed during the rst months of the childs life. The children are from middle class, well-educated families. Data collection started when the children were beginning to combine two words.5 Approximately every three to ve months three one-hour sessions were held with Y and BT. All three sessions took place in the course of 7 to 10 days. The experimenter was instructed to interact with the child in a natural way, focusing on activities that encouraged conversation. Child and experimenter played together, had meals, read books and engaged in other home activities. Specic manipulations were not attempted. The sessions were audio-recorded and later transcribed by the person who collected the data. The same experimenter saw both children for the whole period of data collection. The analysis below is based on 18 hours of recorded conversations with Y and BT, divided into ve periods covering the duration of the study. Y was 3 ; 95 ; 3 during data collection and BT was 4;25 ;8. MLU for both children was 1.82.8. MLU was calculated according to the system adapted for Hebrew by Dromi & Berman (1982) and revised by Levy (1995). In addition to matching on MLU at the beginning and end of the study, the children were matched on proportion of utterances with MLU longer than 5. This procedure is necessary to control for potential biases that are inherent in central measures such as the mean. For Y the proportion of utterances of length 5 and above was 7.716.8%. For BT this proportion was 618.7%. Table 1 gives ages, number of analysed utterances, MLU and percent of utterances of length 5 and above for Y and BT for the duration of the study. Twenty TD children, aged 2; 02; 4, MLU 2.22.8 (mean=2:6 ; S.D.=0: 16) served as controls. The range of MLU and of percent of utterances of length 5 and above in the controls was the same as it was for Y and BT during the periods of the study. The range of utterances of
[5] In fact, the experimenter visited the children for nearly 6 months prior to the rst session reported here. During that time there were no word combinations and the children spoke only single words. These sessions are therefore not included in the current report.

295

LEVY TABLE

1. Age, number of analysed utterances, MLU and % of utterances of length 5 and above for Y and BT
Number of analysed utterances BT 3; 10.20 4; 2.04 4; 7.01 5; 0 5; 4.02 Y 601 532 509 486 604 BT 652 407 642 646 438 % utterances 5 and longer Y 7.7 13.5 13.9 14.4 16.8 BT 6 9.8 11.2 10.7 18.7

AGE Period I II III IV V Y 4; 3.26 4; 8.05 4; 10.28 5; 3.16 5; 8.03

MLU Y 1.8 2.4 2.4 2.5 2.8 BT 1.8 2.2 2.1 2.3 2.8

length 5 and above in the controls was 8.415%. This percentage is higher than that reported in Dromi & Berman (1986) for their sample of Hebrewspeaking two-year-olds. Note that the lowest MLU in the control children is higher than MLU in Y and BTs earliest recordings. Thus, the controls are matched to Y and BT on MLU for period IIV, yet not for period I. This is due to the fact that we could not nd children with MLU under 2 who could also be matched to Y and BT on percent of sentences longer than 5. Similar to Y and BT the children came from middle-class, educated homes. Each control child was seen twice on consecutive days, for a total of 11:30 hours. A similar procedure for obtaining speech samples to those described above was applied. Altogether 3523 utterances were analysed for the controls. Although the control data was cross-sectional rather than longitudinal, it was drawn from a group of children whose MLU and percent of sentences longer than 5 covered most of the period of data collection from Y and BT. Furthermore, the comparisons involved the distribution around the mean rather than a single mean value (see next section) and thus spanned most of the developmental phase studied in Y and BT. Coding and analysis The complete corpora were coded twice, by two independent coders. Child utterances were coded according to the CHILDES system (MacWhinney, 1995), supplemented by a special coding system for Hebrew texts (Levy, 1995). Excluded from the coding were imitations, repetitions, game-like utterances as well as utterances in which there were problems of articulation. The les were later compared and disagreements over coding were resolved by discussion. Utterances for which agreement could not be reached were discarded from the analysis. Those constituted 0.8% of Ys 296

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT

and 0.6% of BTs entire corpus. Thus the analysis presented below is based on instances for which there were complete agreement between the coders. Coding referred to correct usage as well as to errors and concerned aspects of morphology, syntax, lexical choices and pragmatics. The coding of grammatical errors is problematic because of the nonstandard use that may be found in colloquial speech and the central role that extralinguistic context plays in the interpetation of utterances within discourse. Since the level of education of the childrens parents was comparable to that of the researchers who did the coding, the coders standards of colloquial speech were adopted as the baseline for coding decisions. The errors found in the childrens productions were divided into three main categories : syntactic, morphological and meaning. This broad division oered a rst-pass impression of the childrens performance, although dividing errors into these three categories involved some controversial decisions since certain errors related to more than just one linguistic aspect. Syntactic errors related to diverse syntactic phenomena such as subordinate constructions and the relevant functors, agreement, use of accusative marker /et/, prepositions and ungrammatical subject omissions which are cases of omissions of third person as well as of subject pronouns with present tense verbs.6 Errors of morphology included formal errors involving root consonants, word patterns and inections. Recall that in Hebrew one can often distinguish formal morphological from semantic errors. Errors of meaning related to lexical, semantic and pragmatic aspects of utterances. For example, errors in word choice, use of incompatible terms and problems of reference were considered as tapping lexical-semantic problems. Utterances in which the child used a negative term and immediately contradicted it by using a positive term that reversed the meaning of the statement, or instances in which reference could not be determined from what was actually said were coded as meaning problems. Errors of gender marking on animate nouns, inappropriate marking of tense, person and number, and lack of denite marker when deniteness was required, were likewise counted among the meaning errors. Errors of tense, number, person and gender were coded as meaning errors when the focus was on the correct marking of the semantic concept rather than on agreement or on problems of morphological form. As for the specic variables, those were chosen from among the variables studied in TD children, as summarized in the section on Hebrew. Based on that, the following 12 variables were investigated : error types and percent of use of Hebrew verbless clauses, percentage of complex sentences, correct
[6] Hebrew has a mixed pattern of subject omission. While subjects can be freely omitted in 1st and 2nd person past and future, omission is ungrammatical in 3rd person past and future as well as in all persons in the present tense.

297

LEVY

syntactic agreement, correct use of accusative marker, errors in gender marking on inanimate nouns, errors of gender on animate nouns, use of past tense verbs, errors of word choice and pragmatic infelicity.7 While gender marking on animate nouns, errors of word-choice and pragmatic infelicity as well as the combined count of meaning errors code for semantic and pragmatic aspects of the childrens speech, the other variables are more clearly formal-grammatical. Note that the specic variables also count among the errors in the general categories : the syntactic, morphological and semantic errors. However, the overlap in the count of errors that result from this is small, since the broader categories include many more errors of the relevant types. Percentages were calculated in the following way: percent of the dierent error types was calculated out of the total number of errors committed by the child ; percent of verbless and complex clauses were calculated out of the total number of coded utterances for each child. Percent of use of accusative marker and of agreement were calculated relative to obligatory contexts of usage. That is, use of /et/ was calculated relative to the number of utterances in which an accusative marker was required, while agreement was calculated relative to instances of noun (or pronoun)verb combinations.8 Errors of gender on animate and inanimate nouns were calculated out of the total number of errors of morphology. Percent of past tense was calculated
[7] Pragmatic infelicity relates to cases in which the childs choice of words or construction was such that, had the GENERAL CONTEXT been dierent, that same word or construction would have been appropriate. Discourse exchanges in which the child does not respond to the investigators questions and changes the topic inappropriately, exchanges in which the childs response indicates lack of understanding of the requirements of the conversation are likewise coded as pragmatically inappropriate. The following are examples of pragmatically appropriate and inappropriate uses : Ch : ve-ima ve-aba yiknu lanu and-mom and-dad will buy-pl.3rd to-us And mom and dad will buy us Res : ma ? what? Omission of the direct object results in a grammatical, yet, referentially opaque and thus pragmatically unacceptable sentence. The following is an example of a pragmatically appropriate exchange : Ch : tavi li uga kazot bring-3rd, fem.sg me cake like-this get me a piece of this cake Res : ze nigmar, mictaeret Its nished, sorry Ch : az kazoti then like-this, fem.sg. so, like this one [8] In Hebrew, adjectives and certain prepositions likewise agree with the head noun. However, there were no such instances in the childrens productions.

298

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT

3.1 Y BT
2.3

2.7

1.9 1.5 1 2 3 4 5
Fig. 1. MLU for Y and BT for periods IV.

out of the total number of verbs used by the child. Errors of word-choice and errors of pragmatics were calculated out of the total number of errors of meaning committed by the child. The mean and standard deviation (S.D.) of the control group provide measures of the performance and the variance seen in TD children from period II through V. Y and BT were considered similar to the controls with respect to a given variable if their achievements were within one S.D. from the normal mean. In such cases Y and BTs performance was marked y. If Y and BTs error rates were above one S.D. or their correct performance was below one S.D., they were considered below the level of the controls, marked as x. If their performance was above one S.D. and their error rates were below one S.D., they were considered better than the average normal child and their performance was marked +. A similar analysis was adopted in Levy et al. (2000) in which language in children with various neurodevelopmental syndromes was compared to the prole of TD controls. Conti-Ramsden (1998) and Johnson & Carey (1998) use a comparable approach in their respective studies. Note that our focus is on a childs whole prole and the way it compares with the proles of the controls, rather than on performance on individual variables. By selecting a range of possible values around the group mean we are picking up the variance across TD children within the same language levels as Y and BT throughout the period of the study, correcting for the fact that data from the controls was cross-sectional, rather than longitudinal. Findings Growth in MLU for BT and for Y over the period of the study is presented in Figure 1. Both children show an increase in MLU of about 1 point on average during the 18 months of the study, reaching MLU 2.8 in 299

LEVY T A B L E 2. Language proles of Y measured on 12 linguistic variables and their status relative to the mean performance of typically developing controls

Periods Variables Global measures Syntactic errors Morphology errors Specic measures Verbless clauses Complex sentences Past tense verbs Correct accusative /et/ Gender errors animates Agreement errors Controls mean %
S.D.=9.56 S.D.=8.64 S.D.=6.8 S.D.=4.9 S.D.=7.65 S.D.=2.2 S.D.=2.3 S.D.=4.04

29.5

13.6

12.3 5.5

21.4 6.6

3.1

8.5

22.9 y 20.5 y 14.1 y 4.2 y 3 x 6.1 y 0 + 16.7 x 56.6 y 7 y 2.4 + 17.1 x

17.9 + 12.8 y 6.6 y 3.3 y 30 + 11.1 x 0.5 + 0 + 69.4 y 5.4 y 10.2 x 14.3 x

17.9 + 11.9 y 6.2 y 5.5 y 21 y 11.8 x 0.4 + 0 + 70.2 x 4.9 + 11.3 x 11.0 y

17.9 + 11.0 y 6.0 y 7.7 y 13 x 12.5 x 0.4 + 0 + 71.1 x 4.4 + 12.4 x 7.8 y

31.5 y 8.5 y 6.8 y 13.5 + 16 y 3.3 + 1.5 y 7.9 y 60.1 y 0.7 + 9.7 x 12.1 y

Global measures Specic measures

Meaning errors Gender errors inanimates Word choice errors Pragmatic errors

S.D.=11.2 S.D.=5.4 S.D.=1.2 S.D.=3.1

58.2 10.6 3.7

8.9

y, within one S.D. from the mean. x, worse than the controls (below/above one S.D. from the mean of TD controls). +, better than the controls (below/above one S.D. from the mean of TD controls).

the nal session. This level of MLU is achieved by TD children in the rst half of their third year, in other words, about 23 years sooner than Y and BT. Note that MLU for the rst period of the study for both Y and BT is 1.8, which is outside the range of the MLU of the TD controls. A comparison can therefore be drawn between Y and BT and the TD controls in relation to four out of the ve periods of the study. Table 2 gives the performance of Y on the 12 variables for each of the ve periods of the study along with the means and S.D. of the control group. Table 3 gives the scores achieved by BT. Appendix 1 gives raw data for both children. A dividing line is drawn in Tables 2 and 3, between the eight grammatical variables and the last four variables, which are primarily 300

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT TABLE

3. Language proles of BT measured on 12 linguistic variables and their status relative to the mean performance of typically developing controls
Periods Variables Controls mean %
S.D.=9.56 S.D.=8.64 S.D.=6.8 S.D.=4.9 S.D.=7.65 S.D.=2.2 S.D.=2.3 S.D.=4.04

Global measures

Syntactic errors Morphology errors

29.5

13.6 12.3

Specic measures

Verbless clauses Complex sentences Past tense verbs Correct accusative /et/ Gender errors animates Agreement errors

5.5

21.4

6.6

3.1

8.5

15.9 + 12.7 y 13.9 y 2.4 y 24 y 11.8 x 0.5 + 0 + 71.3 x 2.9 + 11.5 x 15.3 x

40.4 x 27.9 x 5.1 + 3.1 y 22 y 8.7 y 0 + 7.7 y 31.7 + 7.3 y 4.8 y 5.8 y

17.4 + 13.2 y 8.1 y 4.2 y 13 x 10.5 x 0 + 3.9 + 69.4 y 2.6 + 15.6 x 15 x

18.5 + 10.2 y 6.3 y 5 y 17 y 0 + 0.4 + 0 + 71.4 x 3.2 + 17 x 12.1 y

19.6 + 10.1 y 6.9 y 10.8 + 26 y 0 + 0.9 y 10.2 y 70.3 x 1.8 + 15.9 x 12.3 y

Global measures Specic measures

Meaning errors Gender errors inanimates Word choice errors Pragmatic errors

S.D.=11.2 S.D.=5.4 S.D.=1.2 S.D.=3.9

58.2

10.6

3.7

8.9

y, within one S.D. from the mean. x, worse than the controls (below/above one S.D. from the mean of TD controls). +, better than the controls (below/above one S.D. from the mean of TD controls).

semantic. The latter are likely to be aected by the childrens lower than normal IQ. Since period 1 is outside the range of MLU of the controls, the comparison will focus on periods IIV with occasional reference to period I. For periods I through IV, Y and BT have a similar percent of verbless clauses as well as a similar percent of complex clauses as do the controls. This suggests that the combined measure of MLU along with percent of sentences longer than MLU 5 is not just measuring length but in fact captures knowledge of syntax as it is reected in the use of sentence types. In period V complex clauses in both children EXCEED the percent seen in the controls. A consideration of the eight grammatical variables in Table 2 reveal + or x signs for Y in 50% of the cells. Table 2 shows 47.5% of the cells for 301

TABLE

4. Performance of children with ND on 13 linguistic variables and their status relative to the mean performance ( +/x S.D.) of typically-developing controls
Children with ND Variables Controls mean %
S.D.=9.5 S.D.=8.6 S.D.=6.8 S.D.=4.9 S.D.=7.6

Mi

Av

Ta

Si

Global measures

Syntactic errors Morpho errors

29.5

13.6

Specic measures

Verbless clauses Complex sentences Past tense verbs

12.3 5.5

21.4

Correct accusative /et/ Gender errors animates Agreement errors

93.4 3.1

S.D.=12 S.D.=2.3 S.D.=4.0

91.5

24 y 7 y 23.4 x 1.4 y 19 y 100 y 4.7 y 100 + 69 y 28.6 x 1.5 + 24 x

20 y 19 y 12 y 3.1 y 17 y 100 y 0 y 100 + 60 y 10 y 3.4 y 16 x

20 y 24 x 18.8 y 2.3 y 20 y 92.3 y 5.6 x 88.2 y 56 y 5.6 y 3.7 y 13 x

37 y 8 y 15 y 7.6 y 21 y 85.7 y 0 y 100 + 55 y 6.7 y 7.4 x 17.2 x

25 y 15 y 5.4 y 2.1 y 11 x 90 y 0 y 100 + 59 y 0.8 + 1.2 + 8.5 y

23 y 17 y 13 y 3.7 y 8 x 86.7 y 0 y 100 + 60 y 15.4 y 3.5 y 4 +

39 y 15 y 16.2 y 14 + 21 y 100 y 2 y 93 y 47 y 21.6 x 2.6 y 3.3 +

25 y 14 y 33.2 x 3.4 y 38 y 85.7 y 3.6 y 82.4 x 61 y 17.8 x 4.5 y 8.2 y

28 y 7 y 16.2 y 3.3 y 19 y 0 y 1.5 y 1.5 x 65 y 7.7 y 8 y 8 x

27 y 14 y 13.9 y 13.1 + 20 y 8.6 y 1.4 y 3.5 y 59 y 14.3 y 6.3 y 6.3 x

27 y 14 y 30 y 4 y 25 y 0 y 2.7 y 1.5 x 59 y 16.2 y 3.1 y 3.1 y

LEVY

302
Global measures Specic measures

Meaning errors Gender errors inanimates Word choice errors Pragmatic errors

58.2
S.D.=11 S.D.=5.4 S.D.=1.2 S.D.=3.1

10.6 3.7

8.9

y, within one S.D. from the mean. x, worse than the controls (below/above one S.D. from the mean of TD controls). +, better than the controls (below/above one S.D. from the mean of TD controls).

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT

BT with + or x signs. That is, the grammatical proles of Y and BT dier from proles of language matched, TD controls on approximately 50 % of the variables points. Interestingly, 37.5% and 32.5% of the cells respectively, are marked as +, that is, Y and BT were doing ABOVE one S.D. of the performance of the TD children in over one third of the cases examined. Note that the +/x signs are not conned to certain proles. However, both children do not have a single x sign in the nal period of the study in any of the grammatical variables, and BT does not have x signs in the period IV either. Within the 8 grammatical variables, accusative /et/, gender of inanimate nouns and agreement are strictly formal and do not involve semantics to the extent that, for example, marking past tense does. Throughout the period of the study the children are either within one S.D. or above the performance of the controls on those variables, except Y whose errors on accusative /et/ exceed one S.D., during periods IIIV. The dierences between the proles of Y and BT and the controls were recalculated relative to an interval of 11 2 S.D. This interval is more traditionally used to dene non-normative performance (Leonard, 1998). For Y, 22.5% of the cells are still outside the normal range, with 10% being above the performance of the controls. For BT too, 22.5% of the cells are outside the normal range, with 12.5% above the performance of the controls and 10 % under. How does the developmental course of Y and BT compare with proles of children with other neurodevelopmental disorders (ND) ? Table 4 gives individual linguistic proles of children with a variety of neurodevelopmental disorders and compares them to the proles of the TD controls. The data in Table 4 reproduces in part Levy et al. (2000) ndings, to which three more subjects were added (Ta, R and Si). Altogether 11 proles are presented. Appendix 2 gives a summary of the childrens neurological conditions, their IQ levels, their MLU and the number of analysed utterances. Individual proles are presented and there is no attempt to average over the group. The range of MLU is 2.22.9 and percent of utterances of MLU 5 and above is 6.817.3%. Thus the children with ND match Y and BT in periods II through V. Similar to Tables 2 and 3, Table 4 divides between the eight grammatical variables and the four semantic ones. However, unlike Y and BT, when the rst eight variables are considered the performance of the children with ND, for its most part, is within one S.D. of the mean of the controls. Of the grammatical variables 19% received either + or x for the 1 S.D. criterion. The scores that depart from the mean of the controls are found all over the table and do not characterize specic children. No more than 8% of the cells that depart from the mean of the TD children received a + sign. Unlike Y and BT, the children with ND are within one S.D. of the controls on all the 303

LEVY

grammatical variables except agreement, which is the only formal variable on which ve children with ND score better than the TD controls. Are the linguistic proles of children with ND more similar to the proles of the TD controls than the linguistic proles of Y and BT? In order to draw such a comparison, each child with ND was given a score which is the sum of the cells for which there is a y sign for that child. A maximum score a child could obtain on the full prole was 12. A maximum score on the grammatical prole was 8. A similar scoring procedure was applied to proles IIV for Y and BT. Their overall score was divided by 4, to obtain the mean number of y across the periods of the study. The scores obtained by the children with ND were compared to the mean scores of Y and BT (MannWhitney, Statistica, release 4.0, Statsoft, Inc). Both comparisons yielded statistically signicant dierences (Full prole, z=2.48; p<0.013 ; Grammatical prole, z=2.36; p<0.018).

SUMMARY AND DISCUSSION

The current work investigated the acquisition of Hebrew in two pre-schoolers with WS. The children were followed for 18 months beginning with the emergence of word combinations. The data consist of conversations with an adult in a naturalistic context. Linguistic proles were drawn for the children at ve points in time, when MLU was 1 ;82;8. MLU increased very slowly during the period of the study. In their fourth and fth year, Y and BT were at an MLU level typical of 2; 02 ;4. These ndings are not surprising since delay in the onset of language as well as slowness in growth of MLU is common in children with WS, although the extent of the delay varies considerably (Capirci et al., 1996; SingerHarris et al., 1997 ; Mervis & Robinson, 2000). Delay is also very common in children with other neurodevelopmental disorders. In fact, it is the hallmark of language development in disordered populations. Y and BT were compared to TD children of equivalent MLU and percent of sentences longer than 5. This was group data collected at one point in time. Note that these data cover the developmental period during which data was collected from Y and BT, except period I, when MLU for both Y and BT was lower than the range seen in the controls. Interestingly, we did not nd TD children with MLU lower than 2, yet with percent of sentences longer than 5 that will parallel that of Y and BT. This was the rst indication that MLU may not reect the same facts about the language of Y and BT as it does in the case of TD children. The comparison drawn between Y and BT and the controls takes into consideration the variability seen in normal development by looking at a group of language-matched children and considering the distribution around the mean, and by asking whether at any point during that developmental 304

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT

phase (as measured by MLU and percent of sentences longer than 5), the proles of Y and BT are similar to proles of TD children. We are interested in the entire prole rather than in individual variables. The ndings suggest that in periods IIV the proles of Y and BT diered on average from the proles of the controls in 50% of the variables measured at various points in time. Recalculating the dierences with respect to 11 2 S.D. resulted in a dierence in performance on 22% of the measured variables. The dierences are outstanding especially in view of the far greater similarity to the normal that is seen in the proles of children with ND of equivalent MLU, and the statistically signicant dierences between the proles of Y and BT and proles of children with other ND, overall, as well as with respect to the grammatical variables. Does growth in MLU predict grammatical development in Y and BT? In the rst recordings, MLU in Y and BT was lower than that of the controls. However, percent of verbless clauses and complex clauses was within the normal range for Y and BT during that rst period, and performance on most of the variables was either within or above the range seen in the controls. While the dierences in performance between the children with WS and the controls in periods IIV are in both directions, the children outperform the normal controls on 35 % of the variables, whereas they are poorer on just 15 % of the variables. Marking gender on inanimate nouns and agreement, which are formal-grammatical variables, are among the ones that Y and BT perform the best on. These data suggest that MLU may underestimate the childrens grammatical development. This is particularly suggestive given how similar to the normal are the proles of children with other ND. This conclusion does not hold, however, for the linguistic variables that are semantic-conceptual where higher than normal error rates have been observed. This is not surprising given the childrens general cognitive handicaps. The increased number of errors in word choice is nevertheless unexpected in view of the reported good vocabularies in children with WS (Mervis & Robinson, 2000). It is possible that these conicting results are due to the dierent methodologies used. Mervis & Robinson are based on parental reports, which, while noting usage, are presumably less sensitive to the accuracy and appropriateness in context of the words that the children use. When correct usage is the focus of the analysis, the data show that children with WS perform less well than language matched normal controls. Did our study uncover atypical trajectories en route to mature linguistic performance, as predicted by Karmilo-Smith (1998)? Throughout the period of data collection, no unusual error types of the kind that are never encountered in TD children were committed by either Y or BT. The longitudinal follow up revealed asynchrony between MLU and grammatical complexity in the rst 2 years of the development of grammar. MLU, 305

LEVY

proportion of shorter vs. longer utterances in discourse and amount of errors and correct use of specic variables were not synchronized in Y and BT in the same way that they were in TD children. A possible interpretation of asynchrony is that it signals deviance. However, issues of timing that are related to the developmental pace of various components of language are so common in neurodevelopmental disorders (Tager-Flusberg, 1999), and asychrony may be too weak a case to argue for atypicality. For example, disorders of timing are most clearly seen in the overall delay in the onset of language which is common in WS as well as in other neurodevelopmental disorders (Singer-Harris et al., 1997 ; Levy et al., 2000). This phenomenon is not considered sucient evidence for a deviant trajectory. I suggest that the asynchrony that has been documented in the current study with respect to individual variables is of the same nature, involving a unique developmental pace rather than a deviant trajectory. Support for this view comes from the commonality of atypical timing in the language of WS at various developmental points. Thomas et al. (2001) report an overall delay in the acquisition of past tense which is seen in regular as well as irregular verbs. Paterson et al. (1999) study showed a lack of advantage for toddlers with WS in word learning over MLU matched children with Down syndrome who are typically delayed in word acquisition. These unexpected ndings reect an issue of timing of vocabulary development relative to the expected WS linguistic phenotype. Finally, Jarrold, Baddeley & Hewes (1998) argued that verbal abilities in WS develop at a higher rate than visual-perceptual abilities and this dierence in rate of development predicted that dierences in performance between those abilities will be seen in older persons and in particular, in individuals who have reached a higher level of verbal abilities. Dierences among various sub-tests are minimal in individuals with WS who have low levels of achievements. This conclusion has recently received support in a longitudinal study of the individuals in Jarrold et al.s (1998) original study (Jarrold, Baddeley, Hewes & Phillips, 2001).

REFERENCES Bellugi, U., Bihrle, A., Neville, H., Doherty, S. & Jernigan, T. (1992). Language, cognition and brain organization in neuro-developmental disorders. In M. Gunnar & C. Nelson (eds), Developmental behavioral neuroscience. Hillsdale, NJ : Erlbaum. Bellugi, U., Marks, S., Bihrle, A. & Sabo, H. (1993). Dissociations between language and cognitive functions in Williams syndrome. In D. Bishop & K. Mogford (eds), Language development in exceptional circumstances. Hillsdale, NJ : Erlbaum. Berman, R. A. (1985). The acquisition of Hebrew. In D. I. Slobin (ed.), The cross-linguistic study of language acquisition. Hillsdale, NJ : Erlbaum. Berman, R. A. (1994). Developmental perspectives on transitivity : a conuence of cues. In Y. Levy (ed.), Other children, other languages. Hillsdale, NJ : Erlbaum.

306

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT

Capirci, O., Sabbadini, L. & Volterra, V. (1996). Language development in Williams syndrome : a case study. Cognitive Neuropsychology 13(7), 101739. Clahsen, H. & Almazan, M. (1998). Syntax and morphology in Williams syndrome. Cognition 68, 16798. Clahsen, H. & Almazan, M. (2001). Compounding and inection in language impairment : evidence form Williams syndrome (and SLI). Lingua 111, 32352. Clahsen, H. & Temple, C. (2003). Words and rules in Williams syndrome and SLI. In Y. Levy & J. Schaeer (eds), Language competence across populations : toward a denition of SLI. Erlbaum. Conti-Ramsden, G. (1998). What is the nature of specic language impairment? Is SLI really specic? In A. Baker, M. Beers, G. Bol, J. de Jong & G. Leemans (eds), Child language disorders in a cross-linguistic perspective. ASCLD, 6, Universiteit van Amsterdam, nr. 71. Dromi, E. & Berman, R. A. (1982). A morphemic measure of early language development : data from Modern Hebrew. Journal of Child Language 9, 40324. Dromi, E. & Berman, R. A. (1986). Language general and language specic in developing syntax. Journal of Child Language 14, 37187. Gosch, A., Stading, G. & Pankau, R. (1994). Linguistic abilities in children with WilliamsBeuren Syndrome. American Journal of Medical Generics 52, 2916. Hyams, N. & Wexler, K. (1993). On the grammatical basis of null subject in child language. Linguistic Inquiry 24, 42159. Jarrold, C., Baddeley, A. & Hewes, A. K. (1998). Verbal and nonverbal abilities in the Williams syndrome phenotype : evidence for diverging developmental trajectories. Journal of Child Psychology and Psychiatry 39, 51123. Jarrold, C., Baddeley, A., Hewes, A. K. & Phillips, C. (2001). A longitudinal assessment of diverging verbal and non-verbal abilities in the Williams syndrome phenotype. Cortex 37(3), 42332. Johnson, S. C. & Carey, S. (1998). Knowledge enrichment and conceptual change in folk biology : evidence from Williams syndrome. Cognitive Psychology 37(2), 156200. Karmilo-Smith, A. (1998). Development itself is the key to understanding developmental disorders. Trends in the Cognitive Sciences 2, 38998. Karmilo-Smith, A., Scerif, G. & Thomas, M. (2002). Dierent approaches to relating genotype to phenotype in developmental disorders. Developmental Psychobiology 40(3), 31122. Karmilo-Smith, A., Grant, J., Berthoud, I., Davies, M., Howlin, P. & Udwin, O. (1997). Language and Williams syndrome : how intact is intact ? Child Development 68, 27490. Laing, E., Butterworth, G., Ansari, D., Gsodl, M., Longhi, E., Panagiotaki, G., Paterson, S. & Karmilo-Smith, A. (2002). Atypical development of language and social communication in toddlers with Williams syndrome. Developmental Science 5(2), 23346. Leonard, L. B. (1998). Children with specic language impairment. Cambridge, MA : MIT Press. Levy, Y. (1988). On the early learning of formal grammatical systems : evidence from studies of the acquisition of gender and countability. Journal of Child Language 15, 17987. Levy, Y. (1995). Coding manual for Hebrew texts-Revised. Publications in Developmental Psychology 2. Levin Institute, The Hebrew University. Jerusalem. Levy, Y. (2003). Early language in congenital disorders and the notion of brain plasticity. In Y. Levy & J. Scheaer (eds), Language competence across populations toward a denition of specic language impairment. Hillsdale, NJ : Erlbaum. Levy, Y., Tennebaum, A. & Ornoy, A. (2000). Spontaneous language of children with specic neurological syndromes. Journal of Speech, Language and Hearing Research 43. Levy, Y. & Bechar, T. (2003). Cognitive, lexical and morphological prole of Israeli adolescents with Williams syndrome. Cortex 39. Levy, Y. & Hermon, S. (in press). Grammar in Hebrew-speaking adolescents with WS. Developmental Neuropsychology (Special Issue on Williams syndrome). MacWhinney, B. (1995). The CHILDES Project. Hillsdale, NJ : Erlbaum.

307

LEVY

McKusick, V. (1988). Mendelian inheritance in man : catalog of autosomal dominant, autosomal recessive and X-linked phenotypes. Baltimore : Johns Hopkins University Press. Mervis, C. B. & Bertrand, J. (1997). Relations between cognition and language : a developmental perspective. In L. B. Adamson & M. A. Romski (eds), Research on communication and language disorders : contributions to theories of language development. NY : Brooks, 1994. Mervis, C. B., Morris, C. A., Bertrand, J. & Robinson, B. F. (1999). Willliams syndrome : ndings from an integrated program of research. In H. Tager-Flsuberg (ed.), Neurodevelopmental Disorders. MIT Press. Mervis, C. & Robinson, B. (1999). Methodological issues in cross-syndrome comparisons : matching procedures, sensitivity (Se) and specicity (Sp). Invited commentary, Child Development Monograph. Mervis, C. B. & Robinson, B. F. (2000). Expressive vocabulary ability in toddlers with Williams syndrome or Down syndrome : a comparison. Developmental Neuropsychology 17(1), 11126. Mulford, R. & Morgan, J. L. (1983). On learning to assign gender to new nouns in Icelandic. Paper presented at the Ninth Annual Minnesota Regional Conference on Language and Linguistics, Minneapolis, May 1314, 1983. Osborne, L. R., Li, M., Pober, B., Chitayat, D., Bodurtha, J., Mandel, A., Costa, T., Grebe, T., Cox, S., Tsui, L.-C. & Scherer, S. (2001). A 1.5 million-base pair inversion polymorphism in families with Williams-Beuren syndrome. Nature Genetics 29, November 2001. Paterson, S. J., Brown, J. H., Gsodl, M. K., Johnson, M. H. & Karmilo-Smtih, A. (1999). Cognitive modularity and genetic disorders. Science 286, 17 Dec. 1999. Pezzini, G., Vicari, S., Voltera, V., Milani, L. & Ossella, M. T. (1999). Children with Williams syndrome : is there a single neuropsychological prole? Developmental Neuropsychology 15(1), 14155. Rice, M. L. & Wexler, K. (1996). A phenotype of specic language impairment : extended optional innitives. In M. Rice (ed.), Toward a genetics of language. Hillsdale, NJ : Erlabaum. Singer-Harris, N., Bellugi, U., Bates, E., Jones, W. & Rossen, M. (1997). Contrasting proles of language development in children with Williams and Down syndromes. Developmental Neuropsychology 13(3), 34570. Smoczynska, M. (1985). The acquisition of Polish. In D. I. Slobin (ed.), The crosslinguistic study of language acquisition. Vol. 1. NJ : Erlbaum. Tager-Flusberg, H. (1999). Language development in atypical children. In M. Barrett (ed.), The development of language. Hove, Sussex : Psychology Press. Tassabehji, M., Metcalfe, K., Karmilo-Smith, A., Carette, M. J., Grant, J., Dennis, N., Reardon, W., Splitt, M., Read, A. P. & Donnai, D. (1999). Williams syndrome : use of chromosomal microdeletions as a tool to dissect cognitive and physical phenotypes. American Journal of Human Genetics 64(1), 11825. Thomas, M. S. C., Grant, J., Gsodl, M., Laing, E., Barham, Z., Lakusta, L., Tyler, L. K., Grice, S., Patterson, S. & Karmilo-Smith, A. (2001). Past tense formation in Willliams syndrome. Language and Cognitive Process 16, 14376. Udwin, O. & Dennis, A. (1995). Psychological and behavioral phenotype in genetically determined syndromes : a review of research ndings. In G. OBrien & W. Yule (eds), Behavioral phenotypes. London : MacKeith. Volterra, V., Capirci, O., Pezzini, G., Sabbadini, L. & Vicari, S. (1996). Linguistic abilities in Italian children with Williams syndrome. Cortex 32, 66377. Williams, J. C. P., Baratt-Boyes, B. G. & Lowe, J. B. (1961). Supravalvular aortic stenosis. Circulation 24, 131118. Zukowski, A. (2001). Uncovering grammatical competence in children with Williams syndrome. Unpublished dissertation, Boston University, Graduate School of Arts and Sciences.

308

A LONGITUDINAL STUDY OF LANGUAGE DEVELOPMENT

APPENDIX 1
RAW DATA FOR TABLES 1 AND 2

Table 1 : Raw data for Y


Variables Syntactic errors Morphology errors Verbless clauses Complex sentences Past tense verbs Correct accusative /et/ Gender errors animates Agreement errors Meaning errors Gender errors inanimates Word choice errors Pragmatic errors Y1 47/205 42/205 116/205 27/191 8/191 2/33 2/12 0/68 13/68 5/205 35/205 3/91 Y2 35/196 25/196 136/196 14/212 5/212 4/36 0/20 1/78 12/78 80/196 28/196 39/129 Y3 37/207 25/207 146/207 14/224 8/224 3/22 0/23 1/76 12/76 49/207 28/207 28/136 Y4 39/218 24/218 155/218 14/236 18/236 1/8 0/25 1/74 11/74 17/218 27/218 17/143 Y5 78/248 21/248 149/248 20/296 40/296 1/30 3/38 4/76 2/76 24/248 30/248 31/204

Table 2 : Raw data for BT


Variables Syntactic errors Morphology errors Verbless clauses Complex sentences Past tense verbs Correct accusative /et/ Gender errors animates Agreement errors Meaning errors Gender errors inanimates Word choice errors Pragmatic errors BT 1 25/157 20/157 112/157 35/252 6/252 4/34 0/29 1/53 6/53 18/157 24/157 28/117 BT 2 42/104 29/104 33/104 8/157 5/157 2/23 1/13 0/60 14/60 6/104 6/104 23/104 BT 3 29/167 22/167 116/167 19/235 10/235 2/19 1/26 0/48 6/48 26/167 25/167 16/127 BT 4 38/206 21/206 147/206 19/300 13/300 0/19 0/56 1/41 9/41 35/206 25/206 34/191 BT 5 27/138 14/138 97/138 16/233 25/233 0/7 6/59 2/43 4/43 22/138 17/138 39/150

309

LEVY

APPENDIX 2 T Y P E O F S Y N D R O M E, I Q, M L U, C A A N D S I Z E O F C O R P O R A A N A L Y S E D F O R T H E C H I L D R E N W I T H N D* A, B, and Ta have fragile X syndrome; Mi has Sotos syndrome; M and E have hydrocephalus ; Av and R have left hemisphere infarct ; S and Si have enlarged ventricles and T has left hemiatrophy
child A Mi M B E Av T S Ta R Si syndrome fraX Sotos Hydroc fraX Hydroc LHInfarct Left Hemiatrop EnlargV fraX LHInfarct EnlargV IQ **GCI=48 verbal=23 GCI=50 verbal=24 GCI=63 verbal=35 IQ=70 verbal=70 Leiter=58 Leiter=60 IQ=69 verbal=75 Bayley=74 Bayley=73 IQ=70 Bayley=69 MLU 2.2 2.5 2.5 2.9 2.8 2.3 2.8 2.4 2: 1 2: 7 2: 6 age 4; 8 6; 10 3; 3 3; 5 3; 5 4; 4 3; 6 3; 6 2; 10 3; 4 3 total utterances 383 376 587 365 739 530 536 533 564 433 523

* A, Mi, and M were tested on the McCarthy (1972). B, T and R were tested on the Stanford-Binet (1960). E and Av were tested on the Leiter (1969) and S, Ta and Si were tested on the Bayley (1969). ** Verbal scores for the McCarthy (1972) have a mean of 50 (S.D.=10) while the GCI has a mean of 100 (S.D.=15).

310

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Вам также может понравиться